cross-shelf dynamics in a western boundary current regime: implications ...€¦ · cross-shelf...

Cross-Shelf Dynamics in a Western Boundary Current Regime: Implicationsfor Upwelling

AMANDINE SCHAEFFER, MONINYA ROUGHAN, AND BRADLEY D. MORRIS

Coastal and Regional Oceanography Laboratory, School of Mathematics and Statistics, University of New South Wales,

Sydney, New South Wales, Australia

(Manuscript received 16 September 2012, in final form 18 December 2012)

ABSTRACT

The cross-shelf dynamics up- and downstream of the separation of the South Pacific Ocean’s Western

Boundary Current (WBC) are studied using two years of high-resolution velocity and temperature mea-

surements frommooring arrays. The shelf circulation is dominated by theEastAustralianCurrent (EAC) and

its eddy field, with mean poleward depth-integrated magnitudes on the shelf break of 0.35 and 0.15 m s21 up-

and downstream of the separation point, respectively. The high cross-shelf variability is analyzed though

a momentum budget, showing a dominant geostrophic balance at both locations. Among the secondary

midshelf terms, the bottom stress influence is higher upstream of the separation point while the wind stress is

dominant downstream. This study investigates the response of the velocity and temperature cross-shelf

structure to both wind and EAC intrusions. Despite the deep water (up to 140 m), the response to a dominant

along-shelf wind stress forcing is a classic two-layer Ekman structure. During weak winds, the shelf en-

croachment of the southward current drives an onshore Ekman flow in the bottom boundary layer. Both the

bottom velocity and the resultant bottom cross-shelf temperature gradient are proportional to the magnitude

of the encroaching current, with similar linear regressions up- and downstream of the WBC separation. The

upwelled water is then subducted below the EAC upstream of the separation point. Such current-driven

upwelling is shown to be the dominant driver of cold water uplift in the EAC-dominated region, with sig-

nificant impacts expected on nutrient enrichment and thus on biological productivity.

1. Introduction

Cross-shelf circulation is a key component of the dy-

namics on continental shelves. It influences water strati-

fication, cross-shelf exchange, and mixing or entrainment

of water masses. The dynamics across the shelf control

primary productivity as vertical uplift supplies nutrients

into the euphotic zone. Furthermore, variability in cross-

shelf structure has been shown to aid in either cross-shelf

transport or inshore retention especially during upwelling

(Roughan et al. 2006). Many physical processes interact to

control the complex dynamics in continental shelf regions.

To aid our understanding of these complex interactions,

typically, the continental shelf is divided into different

zones: surfzone, inner, mid-, and outer shelf (or shelf

break). Many studies have focused on the inner shelf,

where the dynamics tend to be primarily wind driven, but

also influenced by stratification and river discharge (Lentz

2001; Fewings et al. 2008; Dzwonkowski et al. 2011a,b). At

the mid- and outer shelf, cross-shelf structure can be more

complicated than 2D wind-driven flow, as the surface and

bottom boundary layers are separated by an interior flow

(Dever 1997; Liu and Weisberg 2005). The large-scale

circulation can then significantly interact with the coastal

dynamics, even driving upwelling through bottom stress

(Oke and Middleton 2000; Roughan and Middleton

2004; Hyun and He 2010; Castelao 2011).

The focus of this study is to examine the mechanisms

that drive the cross-shelf dynamics along the continental

shelf of eastern Australia. In this region the large-scale

circulation is dominated by the East Australian Current

(EAC), which forms the western boundary of the South

Pacific Ocean’s subtropical gyre. It flows poleward along

the coast of easternAustralia transporting heat and biota,

as shown in the typical summer condition of sea surface

temperature (SST) and geostrophic dynamics (Fig. 1a). It

thus has impacts on coastal weather systems, climate, and

the transport and distribution of species. Less known,

however, is the subsurface impact of the current on

Corresponding author address: Amandine Schaeffer, School of

Mathematics and Statistics, University of New South Wales, Syd-

ney, New South Wales 2052, Australia.

E-mail: [email protected]

1042 JOURNAL OF PHYS ICAL OCEANOGRAPHY VOLUME 43

DOI: 10.1175/JPO-D-12-0177.1

� 2013 American Meteorological Society

continental shelf waters. The EAC has been known to

influence cross-shelf processes in a number of ways in-

cluding driving the upwelling of cold nutrient-rich waters

at the shelf break through bottom friction effects (Oke

and Middleton 2000; Roughan and Middleton 2002,

2004). Encroachment of the jet (at times .2 m s21)

onto the continental shelf displaces shelf waters with

warmer oligotrophic water, potentially advecting pro-

ductivity southward.

Downstream of the separation point of the EAC,

mesoscale eddies are shed from the current. Large warm

core (anticyclonic) and smaller cold core (cyclonic) eddies

form regularly at the separation point (Gibbs et al.

2000;Mata et al. 2006;Wilkin and Zhang 2007; Oke and

Griffin 2011). As the EAC separates from the coast it has

been seen to entrain coastal waters with higher nutrient

concentrations and advect these waters offshore into the

eddy field or along the separating front (Roughan et al.

2011). The EAC usually separates from the coast around

318–328S, as evident in Fig. 1a. From themonthly satellite

observations of SST and geostrophic velocities in 2010/11,

theEACappears to extend as far as Sydney (348S) for just2 months out of the 2-yr study period (not shown). Thus,

Sydney is considered to be downstream of the EAC

separation point with eddy-dominated dynamics.

Until recently this dynamic Western Boundary Cur-

rent (WBC) system and its impacts on the continental

shelf have been observed through intermittent process

studies involving a combination of short mooring de-

ployments and hydrographic surveys, upstream (Oke

et al. 2003), downstream (Cresswell 1994; Gibbs et al.

1998), and straddling the EAC separation point (Roughan

and Middleton 2002, 2004). In addition, remote sensing

has played an important role in elucidating the nature of

the dynamic eddy field (Wilkin and Zhang 2007; Everett

et al. 2012) and, more recently, the associated chlorophyll

response to upwelling via data from Sea-viewing Wide

Field-of-view Sensor (SeaWiFS) and Moderate Resolu-

tion Imaging Spectroradiometer (MODIS) satellites (Oke

and Griffin 2011).

In recent years the Australian Integrated Marine Ob-

serving System (IMOS) has been developed to provide

FIG. 1. (a) The SST (three-day composite) and geostrophic currents (reference vector in top left corner of panel) on

7 Jan 2010. The y axis provides latitude (8S) and the x axis provides longitude (8E). Mean depth-averaged current

vectors (reference vector in top left corner of panels) and variance ellipses at (b) Coffs Harbour [CH070 and CH100;

location indicated by box in (a)] and (c) Sydney [ORS065, SYD100, and SYD140; location indicated by box in (a)]

moorings. The mean wind stress from the stations (reference vector in top left corner of panels) is shown in gray. The

coastline and 20-, 100-, 200-, 500-, 1000-, and 2000-m isobaths are shown, with bold dashed lines for 200 and 2000 m.

MAY 2013 S CHAEFFER ET AL . 1043

sustainable observations of Australian coastal waters. In

particular, arrays of moorings were installed up- and

downstream of the EAC separation point along the coast

of eastern Australia (Roughan et al. 2010). This unprec-

edented long-term dataset of water temperature and veloc-

ity is used here to provide a comprehensive investigation

into the cross-shelf dynamics in this WBC regime.

The paper is organized as follows. The next section

describes the observations and processing used. A brief

description of the dynamics in the region is provided in

section 3. The dominant mechanisms driving the cross-

shelf flow are then highlighted through a momentum

balance analysis (section 4) and uncertainties and limi-

tations are addressed (section 6). In section 5, the in-

fluence of the main forcing mechanisms is investigated,

looking independently at the cross-shelf structure of the

flow when the wind stress or the large-scale circulation is

dominating. Particular attention is given to the resultant

cold water intrusions, with a discussion on the extent of

both the physical and biological response up- and down-

stream of the WBC separation.

2. Observations and data processing

The New South Wales (NWS) node of the IMOS was

designed to examine the physical and ecological inter-

actions of the East Australian Current and its eddy field

with coastal waters. To achieve this, seven moorings have

been deployed along the coast in three across-shelf tran-

sects (Fig. 1) off Coffs Harbour, Sydney, and Narooma

(at 308, 348, and 368S, respectively; Roughan et al. 2010;

Roughan and Morris 2011). The first of these moorings

was deployed in mid-2008. In addition, the Ocean Ref-

erence Station off Sydney (ORS065) has been deployed

since 1990 (Wood et al. 2012). In this study, two years of

observations are analyzed using the Coffs Harbour and

Sydney arrays (latitudes listed in Table 1) from January

2010 toDecember 2011. TheCoffs Harbour array consists

of 2 sites over the mid- (70-m isobath) and outer shelf

(100-m isobath), while off Sydney (450 km to the south)

three sites are instrumented on the 65-, 100-, and 140-m

isobaths. The distance from the coast ranges from 2 to

25 km (Fig. 1 and Table 1).

At eachmooring, a bottom-mountedADCPmeasures

the current velocity in 4-m bins (8 m at SYD140). The

water temperature is also measured through the water

columnwith thermistors at 8-m intervals (4 matORS065),

up to 11–24 m below the surface (Table 1) to avoid dam-

age from boat traffic. These temperature data are com-

plemented by pressure sensors at strategic depths in the

water column. All data are recorded every 5 min and

quality controlled through the IMOS toolbox (http://code.

google.com/p/imos-toolbox/).

The temporal coverage of the measurements (Table 1)

highlights gaps in the time series because of instrument

failures or other data losses. Gaps shorter than 24 h were

filled using linear interpolations, while longer gaps were

not considered for the analysis performed. Monthly hy-

drographic data were also collected off Port Hacking

(20 km south of Sydney sites) and Coffs Harbour (since

September 2011) providing salinity measurements along

cross-shelf conductivity–temperature–depth (CTD) tran-

sects. Hourly wind observations were obtained from the

Bureau of Meteorology at the closest and most signif-

icant sites, Coffs Harbour and Kurnell (Wood et al.

2012), wind stress is computed after Gill (1982) and

Wood et al. (2012), then low-pass filtered in the sameway

as the ocean variables.

The 5-min current and temperature time series are

averaged to hourly intervals and are 38-h low-pass fil-

tered to focus on the subtidal variability using the PL64

filter (Rosenfeld 1983). Current measurements are not

possible in the surface and bottom boundary layers be-

cause of limitations in the moored ADCP configuration

(Table 1). Hence, the observations are extrapolated to

the bottom and surface assuming a constant velocity

following Shearman and Lentz (2003), to compute the

cross-shelf full depth transports. The analyses were also

performed using a linear extrapolation to the surface

and bottom but showed little differences. An across- and

along-shelf coordinate system was determined using the

orientation of the principal axis of the depth-averaged

TABLE 1. Mooring information for CH070 (midshelf), CH100 (shelf break), ORS065 (midshelf), SYD100 (midshelf), and SYD140 (shelf

break). The temporal coverage is calculated over a period of two years, from January 2010 to December 2011.

Current Temperature

Name Lat (8S)Water

depth (m)

Distance

offshore (km)

Major axis

orientation (8)Bin depth

(m)

Temporal

coverage (%)

Sensor

depth (m)

Temporal

coverage (%)

CH070 30.27 74 16 7 10–65 82 16–72 100

CH100 30.27 98 25 7 13–89 75 11–96 90

ORS065 33.90 67 2 16 11–61 100 16–66 100

SYD100 33.94 104 10 19 12–96 89 24–102 100

SYD140 33.99 138 19 24 23–127 98 21–137 86


current, summarized in Table 1, with the x and y axes

chosen to be positive for offshore and northeastward

flows, respectively. The choice of this rotation angle is

discussed in section 6.

3. Context

a. Mean flow

The mean depth-averaged current vector and variance

ellipse at each site (computed from the two years of

measurements) shows that the circulation is predomi-

nantly southward along the shelf with higher magnitudes

at Coffs Harbour (0.35 m s21 at CH100) relative to

Sydney (0.15 m s21 at SYD140) (Fig. 1). This is consis-

tent with the prevailing synoptic circulation, with the in-

tense EAC flowing poleward along the coast until around

318S where it partly bifurcates eastward, as shown on the

typical SST and geostrophic velocitymap of the 7 January

2010 (Fig. 1a). The shelf at 348S (Sydney) is then domi-

nated by a resulting weaker EAC flow or the western

branch of large warm core eddies (Godfrey et al. 1980;

Ridgway and Dunn 2003). The mean circulation is also

more intense over the shelf break (CH100 and SYD140),

with the mean southward velocity at the midshelf site

being only 0.21 and 0.08 m s21 (CH070 and ORS065,

respectively). Though the along-shelf current variability

shown with the ellipses is high at all locations it is, rela-

tively, higher off Sydney than off Coffs Harbour when

compared with the average velocity vectors. This is again

a consequence of the large-scale circulation-inducing in-

termittent current reversals; for instance, when cold core

eddies encroach (Oke and Griffin 2011) off Sydney while

upstream (Coffs Harbour), the southward currents are

observed more than 80% of the time. Even with the

dominant along-shelf circulation, the variance ellipses

still show across-shelf variability, which is of significance

to the work presented here.

The mooring arrays up- and downstream of the EAC

separation were used to generate cross-isobath sections

of velocities and temperature (Fig. 2). The temporal

averages over the whole 2-yr period show the vertical

structure of the dynamics. The mean along-shelf velo-

city is highest in the surface layer both up- and down-

stream, with maxima of 0.5 and 0.2 m s21 respectively,

decreasing with depth. The water is overall warmer

(17.88–238C) and more stratified up- than downstream

(14.88–19.98C) by 28–38C. At Coffs Harbour, the influ-

ence of the intruding EAC is also evident in the isotherm

tilting, with the 228C contour roughly coinciding with the

maximum along-shelf current.

The mean across-shelf circulation is onshore at both

Sydney and Coffs Harbour, but weak (less than

0.04 m s21). The only exception is a slightly offshore

current in the bottom boundary layer at SYD100, which

can be attributed to local topographic effects (see the

isobath contours in Fig. 1).

FIG. 2. Average cross-isobath section of (a),(d) along-shelf current, (b),(e) across-shelf current, and (c),(f) temperature for transects at

(top) Coffs Harbour and (bottom) Sydney.


More quantitatively, the cross-shelf heat transport can

be computed for each mooring as r0Cp

Ð 02H uT dz. Here,

u is the cross-shelf velocity, T is the temperature, Cp 53989 J kg21 8C21 is the heat capacity, r0 is the reference

density of the water, andH is the water depth. The EAC

intrusions atCH100 lead to ameanonshore heat transport

more than twice as high as the one measured on the same

isobath downstream (Table 2). Conversely the midshelf

site CH070 is characterized by low heat transport relative

to ORS065, but very high variability with the standard

deviation being 18 times the mean value.

The correlations between velocity components across

the shelf as a function of nominal depth give insight into

the intratransect dynamics (Fig. 3). The adjacentmooring

pairs are highly correlated (.0.7) for all depths when

considering the along-shelf component of velocity, while

Sydney’s furthest sites (ORS065–SYD140) show lower

correlation coefficients (0.5) because of their spatial sep-

aration (Table 1). The corresponding lags are short at the

surface, where the flow is maximum (Fig. 2), increasing

with depth to a maximum of 7 h between SYD100 and

SYD140. The correlation coefficients for across-shelf ve-

locities (Fig. 3b) are lower and more variable, with max-

imum lags of 10 h. Here, the Coffs Harbour sites are still

related but the correlation between the Sydney midshelf

mooring ORS065 and the outer sites is reduced. This

suggests distinct across-shelf dynamics on the Sydney

shelf within a few kilometers of the coast. In contrast to

the along-shelf component, the depth profiles of the

correlations for cross-shelf velocities tend to show a

local maximum close to the bottom, indicating consis-

tent cross-shelf dynamics in the bottom boundary layer

for adjacent moorings.

As the local dynamics are dominated by the EAC or

eddy encroachments on the shelf, the intrusions are

quantified at each mooring site over the 2-yr period

(Fig. 4a). As evidenced in Fig. 1, the along-shelf current

intrusions are more frequent and of higher magnitude up-

stream of the separation point. Upstream, the most intense

EAC encroachment at the shelf break is characterized by

a depth-averaged along-shelf current of21.3 m s21, while

the 25% percentile is20.57 m s21, relative to21.1 m s21

and 20.28 m s21 downstream, respectively. Considering

the intrusion frequency, southward depth-averaged cur-

rents with intensities higher than 0.3 m s21 are observed

34% and 55% of the time at Coffs Harbour (CH070 and

CH100, respectively), while only 9%, 20%, and 22% of

the 2-yr period off Sydney (ORS065, SYD100, and

SYD140, respectively). For section 5, we define current

intrusion into the shelf as y ,20:3m s21, with y being

the depth-averaged along-shelf current at the midshelf

locations (CH070 and SYD100).

b. Wind forcing

The eastern coast of Australia is not characterized by

strong persistent winds. From a 5-yr study of observed

land and overocean wind data from Sydney, Wood et al.

(2012) evidenced three main directions: northward

TABLE 2. Mean (107 W m21) and std dev (107 W m21) of the

cross-shelf heat transport.

CH070 CH100 ORS065 SYD100 SYD140

Mean 21.02 26.88 22.34 22.94 211.02

Std dev 18.13 46.58 7.68 24.46 53.68

Ratio 18 7 3 8 5

FIG. 3. Max correlations and associated time lags between the (a) along- and (b) across-shelf velocity component

from intratransect moorings. Velocity profiles are normalized by the mooring depth in a sigma coordinate system

(1 is top, 0 is bottom). All correlation coefficients are significant at the 95% confidence level.


(downwelling favorable) characterized by the highest

intensities, southwestward (upwelling favorable), and

eastward winds. They also identified the Kurnell land

station as themost relevant site close to Sydney relative

to offshore buoy measurements. Wind observations for

the two years of interest in this paper confirm these

results. The mean measured wind stress is downwelling

favorable, with magnitudes of 0.003 N m22 at Coffs

Harbour and double that off Sydney (Figs. 1b,c).Defining

a weak wind from the low-pass-filtered time series as

jtsj, 0.03 N m22, the wind is weak 67% and 40% of the

time atCoffsHarbour andSydney, respectively. Significant

downwelling conditions (tsy . 0.04 N m22) happen 11%

and 19% of the time at Coffs Harbour and Sydney, re-

spectively (Fig. 4b). Upwelling-favorable winds occur at

both sites (negative along-shelf component in Fig. 4b), with

higher intensity and occurrence downstream of the sepa-

ration point. Using a threshold of20.04 N m22, upwelling-

favorable winds occur 9% and 13% of the time off Coffs

Harbour and Sydney, respectively. Their influence on the

cross-shelf dynamics relative to other forcing mecha-

nisms is investigated in the following sections.

4. Cross-shelf depth-integrated momentumbalance

a. Estimation of terms from observations

Themomentum balance is used to estimate the relative

importance of the different forcing mechanisms driving

the cross-shelf dynamics (Lentz et al. 1999). This analysis

also provides insights into the differences between the

dynamics up- (Coffs Harbour, 308S) and downstream of

the EAC separation point (Sydney, 348S).

The depth-averaged cross-shelf momentum equation

can be written after Oke et al. (2003):

›u

›t2 f y1 u

›u

›x1 y

›u

›y52

1

r0

�›P

›x

�1

tsxr0H

2tbxr0H

, (1)

where (u, y) are the depth-averaged velocities, f is the

Coriolis parameter; tsx and tbx are the surface and bottom

across-shelf stress, respectively; and h›P/›xi is the depth-averaged pressure gradient. Because of limitations in the

mooring array, the nonlinear advection term is limited to

u(›u/›x), since the along-shelf gradient ›u/›y could not

be estimated. This hypothesis is justified by considering

the local dynamics, which do not vary significantly in the

along-shelf direction in regions where the topographic

variations are limited [see Fig. 1a, Oke and Middleton

(2000), and Oke et al. (2003)]. The bottom stress is com-

puted using the linear drag law tbx 5 r0ryb, with the re-

sistance coefficient r5 53 1024 m s21 (Lentz 2001) and

yb the measured bottom along-shelf velocity.

Making the hydrostatic assumption, the pressure gra-

dient term can be computed from the surface or bottom

pressure measurements (Brown et al. 1985). While the

latter is often used when available (Liu and Weisberg

2005; Shearman and Lentz 2003) here the drifts in the

observations of bottom pressure and the bias resulting

from the frequent mooring redeployments (servicing

every 6 weeks on average) led to very noisy values. Thus,

the pressure gradient is computed as

�›P

›x

�5

1

H

ð02H

›P

›xdz5 r0g

›h

›x1

g

H

ð02H

ð0z

›r0

›xdz0 dz ,

(2)

FIG. 4. (a) Histograms of depth-averaged along-shelf currents measured at each mooring site (b) along- and cross-shelf wind stress

observations from Kurnell and Coffs Harbour stations. The location of the stations is shown in Fig. 1.


where g 5 9.81 m s22 is the gravitational acceleration

and r0 is the density anomaly. The sea level gradient

›h/›x is computed from daily satellite altimetry obser-

vations (see section 6 for discussion), and is referred to as

the barotropic pressure gradient, in contrast to the baro-

clinic term based on the density gradients.

All of the terms in the momentum balance, except for

the barotropic pressure gradient, are estimated between

adjacent moorings following the method described in

Liu and Weisberg (2005). New depth-averaged velocity

time series are generated using a weighted sum of both

mooring measurements:

(u, y)5

�u1h11 u2h2h11 h2

,y1h11 y2h2h11 h2

�,

where the subscript numbers correspond to the moor-

ings and h is the local water depth.

The density time series are computed from the tem-

perature observations assuming a constant salinity of 35,

however, a number of quasi-monthly CTD profiles were

also used to justify this assumption.

b. Temporal analysis and standard deviations

The varying importance of the terms in the depth-

averaged momentum balance is examined over the two

years of the study period. Upstream of the separation

point (Coffs Harbour), periods of strong negative Coriolis

forcing correspond to the intrusion of the EAC onto the

shelf ( f , 0 in the Southern Hemisphere) (Figs. 5a,b).

These occur all year round without any clear seasonal

signal [as also observed by Malcolm et al. (2011)]. How-

ever, during summer, they are associated with high neg-

ative baroclinic pressure gradients as the EAC is much

warmer than the shelf waters (March 2010 and January–

February 2011 in Fig. 5a) (Oke et al. 2003). This contrasts

with spring or wintertime (e.g., June, September, and

October 2010, and October and November 2011) when

the baroclinic pressure gradient is weak. The latter time

series computed from themooring temperature assuming

constant salinity is in good agreement with the terms

computed from monthly CTD density profiles off Coffs

Harbour at the end of 2011 (black versus gray dots, re-

spectively; Fig. 5a).

The dominant Coriolis term seems to be balanced by

the sum of both components of the pressure gradient

(Fig. 5b) in geostrophic equilibrium, as suggested by Oke

et al. (2003), although they did not calculate the pressure

term. The secondary terms are shown in Fig. 5c. Both the

wind stress and acceleration terms show high temporal

variability, while the advection and bottom stress ex-

trema are in phase with the Coriolis term. In other words,

FIG. 5. Low-pass-filtered across-shelf momentum balance between Coffs Harbour moorings CH070 and CH100

from January 2010 to December 2011 (m s22). Note that the scales of the y axes differ in the panels. (a) The Coriolis,

baroclinic, and barotropic pressure gradient terms [see Eqs. (1) and (2)]. The dots represent the baroclinic pressure

gradient term from vertically resolved (gray) and constant (black) salinity. (b) The Coriolis term vs the sum of the

pressure gradient terms. (c) The wind stress, bottom stress, local acceleration, and advection terms. Gaps in the time

series indicate periods of data loss.


the EAC intrusions are characterized by relatively im-

portant cross-shelf advection and bottom stress.

The momentum balance downstream of the separa-

tion point shows smaller extrema, both between inner-

(Fig. 6) and outer-shelf mooring pairs (Fig. 7). Note that

the barotropic terms are similar for both because of the

coarse horizontal resolution of the altimetry product

used to compute these pressure gradients (see discussion

section 6). The baroclinic term computed from constant

or vertically resolved salinity are in good agreement

(Fig. 6a) despite the different measurement locations,

the monthly CTD profiles are sampled at Port Hacking,

20 km south of the Sydney moorings (Roughan and

Morris 2011). Here, the negative Coriolis peaks are

shorter than upstream of the separation point and cor-

respond to the western branch of warm core eddies en-

croaching into the shelf (e.g., in January 2010 and 2011,

and inMay, July, and November 2011 in Figs. 6a and 7a)

as confirmed by SST figures (for instance Fig. 1a). The

advection terms are still correlated to theCoriolis forcing,

but the amplitude of the bottom stress is smaller than

upstream, while the wind forcing appears to be one of the

main secondary terms (Figs. 6c and 7c).

This is confirmed by the standard deviations of the

individualmomentum terms listed for eachmooring pair

(Table 3). Of the stress terms, the bottom stress is more

important upstream of the separation point (CH) while

the standard deviation of the wind stress term is higher

downstream (SYD), especially for the inner-shelf pair

where it is the dominant secondary term as the water is

shallow. This is expected to have major impact on ver-

tical movements (see section 5). For all locations, the

strongest terms are Coriolis and the barotropic pressure

gradient, while the standard deviations of the baroclinic

terms are lower. The standard deviations of each of the

terms except wind stress are the largest at Coffs Har-

bour, again highlighting the influence of the EAC on the

cross-shelf dynamics upstream of the separation point.

Downstream, the main difference between the inner-

and outer-shelf pairs is the magnitude of the baroclinic

pressure gradient, which is the largest on the shelf break

relative to the barotropic term. The large standard de-

viation of the residual is attributed to missing data and

uncertainties in some calculations, especially the baro-

tropic pressure gradient term computed from satellite

observations (see section 6). Nevertheless, it is not ex-

pected to have a significant impact on the results of this

study (see section 4c).

c. Correlation and regression analysis

The dominant geostrophic equilibrium at all locations

is evident when comparing the temporal variability and

the magnitude of different terms of the momentum bal-

ance (Table 4). Indeed, both the highest correlation and

regression coefficients are obtained between the Coriolis

and the total pressure gradient term. Including all the

secondary terms does not improve the budget closure as

their magnitude is significantly lower.

FIG. 6. As in Fig. 5, but for the Sydney inner-shelf mooring pair (ORS065-SYD100).


When separating the baroclinic from the barotropic

components of the pressure gradients, the results differ.

As is evident for the data in Table 3, the magnitude of

the baroclinic pressure gradient is low relative to the

Coriolis term, leading to low regression coefficients for

all mooring pairs (0.21–0.33). Nevertheless, these terms

are highly correlated (0.63–0.70) indicating that the main

buoyancy forcing mechanisms are associated with the

synoptic circulation: the EAC upstream and its eddy field

downstream. The geostrophic balance is then mostly

barotropic, with regression slopes closer to 1, while the

weak correlation between the Coriolis terms and the

barotropic pressure gradients can be partly attributed

to the difference in the observation time step (hourly

versus daily because of the lack of observations).

Comparing the different sites, the momentum budget

would appear to be approaching closure, more so at the

Sydney site than the Coffs Harbour site (correlations of

0.72–0.73 and regressions of 0.96–0.99). The weaker sta-

tistics at Coffs Harbour may be because of missing terms

or computational uncertainties (see section 6). Between

the Sydney inner- and outer-shelf locations, the dynamics

FIG. 7. As in Fig. 5, but for the Sydney outer-shelf mooring pair (SYD100-SYD140).

TABLE 3. Std dev (1026 m s22) of across-shelf momentum balance for the CH midshelf (CH070–CH100), SYD inner-shelf

(ORS065–SYD100), and SYD outer-shelf (SYD100–SYD140) pair.

Term

CH midshelf

pair std dev

SYD inner-shelf

pair std dev

SYD outer-shelf

pair std dev

Coriolis 2f y 16.00 11.12 11.63

Barotropic pressure gradient 2g›h

›x13.57 13.13 12.52

Baroclinic pressure gradient 2g

r0H

ð02h

ð0z

›r0

›xdz0 dz 6.07 3.57 5.47

Acceleration›u

›t0.16 0.08 0.13

Advection u›u

›x0.23 0.14 0.17

Wind stresstsxr0H

0.08 0.27 0.16

Bottom stress 2tbxr0H

0.16 0.07 0.07

Residual 13.82 10.43 10.85


appear to be more affected by the buoyancy gradients on

the shelf break.

5. Investigation of forcing mechanisms

Through the depth-integrated cross-shelf momentum

balance, the importance of bottom and surface stress

appeared to vary significantly up- and downstream of the

EAC separation point. This section investigates the re-

spective influence of these forcing mechanisms on the

cross-shelf dynamics with a particular focus on the result-

ing upwelling processes. We separate them by looking

at the circulation when only one of the forcing mech-

anisms is strong, either the along-shelf wind stress, or

the bottom stress resulting from the along-shelf current.

Mean cross sections are generated by averaging the cir-

culation during specific wind or along-shelf current con-

ditions. The interaction of both stresses is also considered

and the frequency of these features over the two years is

discussed.

a. Influence of wind forcing

In the Sydney region,Middleton et al. (1996), Roughan

and Middleton (2002), and Wood et al. (2012) identified

a significant ocean response for oversea wind stress of

around 0.1 N m22, corresponding to 0.04 N m22 lo-

cally over land [Kurnell site; Wood et al. (2012)]. The

latter threshold is used here for along-shelf wind and is

referred to in the following as strong up- or downwelling-

favorable wind when the along-shelf component is dom-

inant (jtsyj. jtsxj). To focus on the response of the oceanto wind forcing only, we also consider weak current

conditions as periods when the midshelf depth-averaged

along-shelf current is greater than20:2m s21. This thresh-

old is chosen to eliminate large-scale southward current

intrusions (EAC or its eddies), while allowing the wind-

driven circulation.

These down- and upwelling proxies are significant,

corresponding to 13% and 10% of the dataset down-

stream and 8% and 2% upstream, respectively (Fig. 8).

At both sites, the cross-shelf dynamics follow Ekman the-

ory (Ekman 1905) when the along-shelf wind stress is the

dominant forcing. In response to a downwelling-favorable

wind, the surface layer is pushed onshore and subducted

at the coast in a deeper offshore flow (Fig. 8, top panels).

At Coffs Harbour, the surface layer is around 30–40 m

deep with onshore velocities up to 0.04 m s21 while the

rest of the water column is characterized by flow in

the opposite directionwithmaximumvelocity close to the

bottomup to 0.06 m s21. Off Sydney, the surface onshore

flow is similar in depth but faster (up to 0.06 m s21) while

the offshore flow is limited to the bottom layer.

The response to a strong upwelling-favorable wind is

shown in Fig. 8 (second row). At both sites, the surface

flow is thinner than for the downwelling response, char-

acterized by an offshore flow of 0.01–0.02 m s21. The

water column below 20-m depth moves onshore to

compensate the depression at the coast with velocities

up to 0.03 m s21 and the isotherms are uplifted relative

to the downwelling scenario.

As expected, the upwelling process is intensified when

the wind stress duration is longer. Off Sydney, the cross-

shelf response to a minimum of 48-h upwelling-favorable

wind stress is shown in Fig. 8 (bottom row, panels on rhs).

Relative to the instantaneous ocean response with the

samewind intensity (see above), the two-layer circulation

is intensified. The near-surface offshore currents are

more intense (up to 0.05 m s21) and extend offshore to

the SYD140 mooring, 20 km away from the coastline.

The water gets colder close to the coast, with an impor-

tant isotherm uplift between the twomidshelf sites (up to

15 m for the 168C isotherm). Unfortunately, over the two

years of observations only three events satisfied the cri-

teria during summertime (one should remember the

limiting weak current criteria as well).

To provide quantification, the cross-shelf response to

wind stress forcing is investigated for different wind in-

tensities. Here, the same proxy y .20.2 m s21 and jtsyj. jtsxj is considered, but the along-shelf wind stress in-

tensity has been binned into 0.02 N m22 intervals, in-

cluding both up- (,0) and downwelling-favorable (.0)

wind stress. Different parameters describing the re-

sponse of the ocean to the forcing are investigated, that is,

the cross-shelf velocity and temperature gradient at two

TABLE 4. Correlation coefficient C and linear regression slope R in the across-shelf momentum balance for the CH midshelf

(CH070–CH100), SYD inner-shelf (ORS065–SYD100), and SYD outer-shelf (SYD100–SYD140) pair against the Coriolis term with zero

lag, for times when all terms are available.

CH midshelf pair SYD inner-shelf pair SYD midshelf pair

C R C R C R

Baroclinic pressure gradient 0.63 0.24 0.64 0.21 0.70 0.33

Barotropic pressure gradient 0.52 0.44 0.64 0.76 0.62 0.67

Baroclinic 1 barotropic pressure gradient 0.65 0.68 0.72 0.97 0.73 0.99

Sum of all terms 0.64 0.68 0.72 0.96 0.73 0.99


FIG. 8. Cross-isobath section of along-shelf velocity y [(first column on lhs) along the Coffs Harbour CH line and (second column on rhs)

Sydney SYD line], cross-shelf velocity uwith temperature contours [along the (second column on lhs) CoffsHarbour line and (first column

on rhs) Sydney line], temporally averaged for different forcing conditions. Conditions are: (top row) downwelling-favorable wind forcing

(.0.04 N m22) during weak southward circulation; (second row) upwelling-favorable wind forcing (.0.04 N m22) during weak south-

ward circulation; (third row) southward current intrusion (.0.3 m s21) during weak wind; (fourth row) simultaneous upwelling-favorable

wind forcing and southward current intrusion; and (bottom row, panels on lhs) strong southward current intrusion (.0.6 m s21) during

weak wind; (bottom row, panels on rhs) 48-h upwelling-favorable wind forcing (.0.04 N m22) during weak southward circulation. The

number of days for each condition and the corresponding percentage of the complete dataset coverage are also specified.


fixed depths, one close to the surface and one in the

bottom layers (Fig. 9a).

Over the two years of measurements, the number of

days when this proxy is satisfied is overall lower up-

stream of the separation point (Coffs Harbour) because

first the wind is weaker (Fig. 4b) and second the EAC

intrusions tend to dominate the circulation off Coffs

Harbour (Fig. 4a). Nonetheless, the major oceanic pat-

terns are similar at both latitudes. The near-surface cross-

shelf velocities show linear trends, with offshore currents

(u . 0) for upwelling-favorable winds and onshore cur-

rents (u , 0) during downwelling-favorable winds. For

the shelf break mooring at Sydney (SYD140), a similar

slope is obtained, but the values are mostly negative, in-

dicating the offshore extent of the wind-driven circula-

tion. The circulation near the bottom is opposed to the

surface circulation in agreement with a classic two-layer

dynamic, but with weaker current magnitudes. At the

ORS065 mooring (located 2 km away from the coast),

the regression line is the closest to the origin for both

surface and bottom cross-shelf velocities, showing an al-

most symmetric response to down-/upwelling processes.

The temperature gradients are predominantly positive,

showing a colder water mass close to the coast. The only

exception is between the Sydney moorings at the 100-

and 140-m isobaths with slightly warmer water at 25-m

depth.While an increase in wind stress does not influence

the bottom isotherms (no significant trend), upwelling-

favorable winds induce a larger near-surface temperature

difference especially between the inshore mooring pair

(SYD100 2 ORS065). The upwelling response although

weak (a gradient of 0.18C km21 corresponds to a 0.88Ctemperature difference between ORS065 and SYD100),

is similar to the findings ofMcClean-Padman and Padman

FIG. 9. (a) Influence of along-shelf wind stress intensity (x axis, discretized in 0.02 N m22 intervals) with a weak southward current

(y .20:2m s21) and (b) influence of along-shelf depth-averaged current intensity (x axis, discretized on 0.1 m s21 intervals), while weak

wind stress (jtsj , 0.03 N m22) for (first two columns on lhs) near-surface and bottom cross-shelf velocity for each mooring, (third and

fourth columns from lhs) near-surface and bottom temperature gradient for eachmooring pair. (fifth column from lhs) The number of days

used to compute each composite value for Coffs Harbour and Sydney moorings are shown, with a lower threshold of 5 days. Each dot is

statistically independent. Because of the lack of boundary layer measurements, the near-surface and bottom are defined as 15- and 25-m

depth for current and temperature respectively, and 5 m above bottom (from the shallower site when a gradient is considered).


(1991) who identified three major wind-driven upwelling

events off Sydney over a 6-yr analysis.

b. Influence of along-shelf current

Typically in this strong WBC regime, the along-shelf

current dominates the cross-shelf dynamics through the

encroachment of the EAC or the western arm of both

cyclonic and anticyclonic eddies. As evidenced in sec-

tion 4, this large-scale circulation impacts the dynamics

of the shelf through Coriolis acceleration, advection,

bottom stress, and buoyancy gradients.

A number of previous studies highlighted the EAC as

amajor driver of current-driven upwelling along the coast,

based on sporadic observations (Cresswell 1994; Gibbs

et al. 1998; Roughan and Middleton 2002, 2004) or mod-

eling (Oke and Middleton 2001; Roughan et al. 2003).

Here we investigate the response of the ocean to the in-

trusion of a strong southward current (y ,20:3m s21

midshelf) using two years of high-resolution observations

up- and downstream of the EAC separation point. We

also consider weak wind conditions jtsj, 0.03 N m22, to

focus solely on the influence of the large-scale dynamics.

These conditions were satisfied 124 and 32 days, repre-

senting 21% and 6% of the complete dataset, up- and

downstream, respectively, over which time the current

and temperature measurements were averaged (Fig. 8,

third row).

The cross-shelf response is more complex than a typ-

ical two-layer wind-driven system. Off Coffs Harbour,

the encroaching EAC is warm (average T of 22–248C)and has an onshore component on the shelf break

(0.04 m s21). In agreement with Ekman theory, the

bottom stress drives an onshore current in the bottom

boundary layer (BBL) (;10–20-m height above bed)

reaching 0.05 m s21 and associated with colder water

(T , 208C). This upwelled water seems then to be ad-

vected offshore at the surface, inducing a surface frontal

convergence zone with the tropical EAC waters. The

offshore extension, variability and vertical movements

induced by this front could not be investigated with only

two moorings, but the impact of a more intense EAC

(y ,20:6m s21) is considered in Fig. 8 (bottom row).

The onshore Ekman current in the BBL is intensified, up

to 0.06 m s21, in agreement with the results of the mod-

eling study by Oke andMiddleton (2000). At the surface,

the frontal zone is more pronounced, and the offshore

flow observed midshelf appears to be subducted under

the intruding EAC on the shelf break.

Downstream, the Ekman geostrophic response to an

intruding southward current (y ,20:3m s21) is also

apparent in the bottom layer on both the shelf break

(SYD140) and the inner site (ORS065). Interestingly,

there is no evidence of such an onshore flow at the

midshelf mooring SYD100. This is likely explained by the

local topography (Fig. 1), as the 100-m isobath shows

some irregularities relative to the other isobaths. Never-

theless, the isotherms are strongly uplifted, both in re-

sponse to the warmer water characterizing the synoptic

circulation and to the bottom onshore flow associated

with colder slope water.

In a similar way to the wind stress, the influence of the

along-shelf current magnitude on the cross-shelf veloc-

ity and temperature gradient close to the surface and in

the bottom layers is investigated (Fig. 9b). The near-

surface cross-shelf velocities are mostly negative and do

not show a linear relationship with the along-shelf cur-

rent intensity, except at the inshore moorings (ORS065

and CH070), where a strong southward current induces

offshore (positive) flow close to the coast. This is con-

sistent with an upwelling feature driven by the bottom

stress, uplifting water along the coast, which is then

driven offshore at the surface by continuity. Indeed, the

bottom velocities are strongly related to the along-shelf

current intensity: northward currents (.0) induce posi-

tive cross-shelf bottom velocities, while the stronger the

EAC or its warm core eddies (WCE) are (southward

currents, ,0), the more intense the onshore bottom

Ekman flow is. This feature is observed for all moorings

except SYD100 where the local topography is believed

to cause the flow to deviate (see discussion above). The

near-bottom temperature gradient at a fixed depth in

response to the along-shelf circulation is surprisingly co-

herent for all moorings, up- and downstream of the sep-

aration point. The regression slope shows coefficients

between 0.27 and 0.33 and zero intercepts ranging from

20.03 to 20.08. This implies that for an along-shelf

southward current of 1 m s21 at midshelf, the bottom

water on the shelf gets colder by 38–3.58C relative to the

same depth 10 km farther offshore.

Close to the surface, a northward flow off Sydney in-

duces a negative temperature gradient at the shelf break.

This corresponds to the encroachment of cold core eddies

leading to warmer water on the shelf (Oke and Griffin

2011). Otherwise the linear regression with negative

slopes at all sites is related to the intrusion of the warm

EAC waters inducing a strong thermal gradient across

the front (Oke et al. 2003). Nevertheless, the slope is

less steep than close to the bottom, indicating a differ-

ent process occurring in the BBL.

The relationship between the bottom onshore flow and

the temperature gradient is emphasized in Fig. 10. In-

cluding all the average values obtained for different

along-shelf current intensities (indicated by the gray-

scale) for all the moorings (Fig. 10a), the R2 value indi-

cates that 62% of the near-bottom composite temperature

gradient is explained by the BBL Ekman flow in


response to bottom stress. The intrusion of the cold

slope water onto the shelf is also evidenced when

comparing the bottom temperature for two adjacent

moorings (Fig. 10b). The bottom temperature at the

midshelf site (CH070) becomes consistent with the

temperature at 100 m on the shelf break (CH100) when

the onshore Ekman flow reaches 0.06 m s21. Down-

stream, the same feature is evident between the ORS065

and SYD100 moorings; however, the intensity of the

bottom flow needs to be higher to completely uplift the

cold water as the depth difference is more important. In

contrast, the R2 value is higher, suggesting a more two-

dimensional process, with a weaker influence of hori-

zontal along-shelf advection through the EAC. Such

slope water intrusions have significant biological impli-

cations as they are expected to carry nutrient-rich water

onto the shelf, enabling primary production.

c. Mixed scenario: Upwelling-favorable wind andcurrent

A number of studies have highlighted the cumulative

effect of the simultaneous occurrence of bottom and

wind stress as a more efficient mechanism for upwelling

(Tranter et al. 1986; Roughan andMiddleton 2002, 2004).

Gibbs and Middleton (1997) suggested that the current-

driven uplift may be a preconditioning for a stronger

wind-driven upwelling. To test this theory, we looked

at the cross-shelf response to a period with concurrent

southward current and upwelling-favorable wind stress

(Fig. 8, fourth row). The forcing magnitudes considered

are the same as used previously: tsy , 20.04 N m22 and

y ,20:3m s21 (see above). Relative to an isolated wind

forcing alone (Fig. 8, second row), the bottom Ekman

flow is more apparent and the isotherms are more up-

lifted both up- and downstream of the separation point.

Relative to a simple along-shelf current forcing (Fig. 8,

third row), the surface offshore flow at the coast is in-

tensified, especially at Coffs Harbour where the EAC

seems to be pushed offshore by the wind-driven Ekman

transport. At the same time, the bottom onshore current

is more intense, with up to 0.10 and 0.05 m s21 at the

Coffs Harbour and Sydney shelf breaks, respectively,

thus twice as high as for the simple current forcing sce-

nario and even more intense than for a stronger current

without wind stress (see above).

6. Uncertainties and limitations

All circulation patterns presented in this study were

defined following an along- and cross-shore coordinate

system based on the principal axis of the depth-averaged

velocity (see section 2). While this is a common practice,

it is still necessary to evaluate to what extent the results

are dependent on this choice. Figure 10a includes results

obtained when the coordinate system is rotated by an

additional 28, either clockwise or counterclockwise. This

FIG. 10. (a) Composite near-bottom temperature gradients from Fig. 9b including all moorings as a function of the

bottom cross-shelf velocity. The corresponding along-shelf depth-averaged current intensity is specified with the

shade. The values of the linear regressionR2 are indicated. The circles and the black line correspond to the coordinate

system described in section 2, while other symbols and dashed lines refer to a coordinate system differing by 628.(b) Bottom temperature difference as a function of the bottom cross-shelf velocity. The R2 values are indicated.


figure was chosen because it includes all mooring data

and is assumed to be themost sensitive to the coordinate

system as it presents the regression between the bottom

temperature gradient and the cross-shelf velocity, for

different along-shelf current intensities. The results ap-

pear to be robust, with significant regressions for all

coordinate systems and fluctuations of R2 values less

than 6%.

Errors in the momentum balance can arise through

various limitations or uncertainties. One of the primary

issues arises from data gaps near the surface. This is

a result of instrument limitations (e.g., ADCP) and the

physical challenges when deploying shelf moorings in an

intense WBC. Furthermore, because of a high volume

of vessel traffic along the coast of eastern Australia, for

security reasons the moorings do not reach the surface

(Table 1). The buoyancy gradient terms presented in

this study were thus computed using only the available

temperature measurements, leading to a probable un-

derestimation of the pressure gradient term. The use of

satellite SST time series for the surface extrapolation of

themooring observations was also tested to compute the

baroclinic term. The standard deviation of the cross-shore

gradients did not change significantly, but the correla-

tions with other terms happened to be reduced by the use

of the daily and cloudy SST dataset. The accuracy of the

barotropic term is a major issue, considering that it was

estimated using altimetry observations, with a low spatial

(0.258) and temporal resolution. The daily sea surface

height (SSH) products were provided by IMOS (Oke and

Sakov 2012), constructed from both coastal tide gauge

observations and altimetry. As the tracks are repeated

only every 10 days, short time-scale dynamics are not

expected to be resolved. An estimation of the charac-

teristic time scales can be obtained by determining the

maximum lag for which the autocorrelation of the time

series is higher than a threshold (for instance 0.7, corre-

sponding to 50% of the explained variability). At the

mooring locations, the characteristic time scale for the

satellite SSH is more than twice as high as that obtained

from in situ depth-averaged velocities (5 days compared

to 1–2 days, respectively). Other uncertainties could

derive from the parameterization of bottom stress as

shown in Lentz (2008). The consequence of the missing

nonlinear terms in the balance is uncertain (Lentz and

Chapman 2004), but in a short-term study the cross-shelf

advective term appeared to be low (Oke and Middleton

2000) and both wave and tidal forcing is excepted to be

negligible in the region. All of these limitations lead to

large residuals (Table 3); however, their magnitude

remains smaller than other terms and the correlation/

regression analysis (Table 4) led to high coefficients such

that we have confidence in the main conclusions.

The wind stress is underestimated using inland rel-

ative to offshore measurements and might be sheltered

by the local orography. The difference in magnitude

for the Sydney station has been taken into account fol-

lowing Wood et al. (2012), but no information is avail-

able for Coffs Harbour station. However, Roughan and

Middleton (2002) and Gibbs et al. (1998) suggest that the

wind stress intensity required for upwelling generation

may be lower than the theoretical value when the water is

already preconditioned by the current, which appears to

be a common feature upstream (see discussion).

7. Discussion and conclusions

The eastern Australian continental shelf circulation is

dominated by along-shelf currents from theWBC (EAC)

flowing poleward and the western arm of warm or cold

core eddies. Nevertheless, the variability of the cross-

shelf circulation is shown to be important on small length

scales (less than 25 km from the coast). This study pres-

ents the first long-term observational analysis of the

cross-shelf dynamics along the east coast of Australia,

both up- and downstream of the EAC separation point.

The mean cross-shelf circulation is weakly onshore at

both sites, with warmer water close to the shelf break,

consistent with the encroachment of the warm EAC or a

WCE.

The forcing mechanisms for a cross-shelf exchange

were identified though an investigation of the momen-

tum balance. Themain dynamical balance is geostrophic

both up- (Coffs Harbour, 308S) and downstream (Syd-

ney, 348S) of the EAC separation, with a dominant bar-

otropic pressure gradient relative to the baroclinic one,

especially at the most inshore moorings. The influence of

the EAC encroachment off Coffs Harbour is also ap-

parent in the secondary terms, through advection, local

acceleration, and bottom stress as modeled by Roughan

et al. (2003). Off Sydney the main secondary driver ap-

pears to be the wind stress.

The major implication of cross-shelf flows is the uplift

of cold and generally nutrient-rich water, which can be

wind- or current-driven (through bottom friction). The

separate response of the cross-shelf velocity and tem-

perature structure to these forcing mechanisms is shown

to be similar up- and downstream of the EAC separation.

During weak large-scale circulation, along-shelf wind

stress drives a classic two-layer circulation, with the cross-

shelf current intensity being linearly dependent on the

wind stress magnitude. As expected, the temperature

structure depends on both the intensity and the duration

of the wind stress.

The encroachment of a southward current onto the

shelf substantially influences the cross-shelf circulation.


The cross-sectional structure shows an onshore bottom

flow in agreement with Ekman theory that uplifts cold

water along the slope. The surface onshore intrusion of

the EAC appears to be limited by the offshore flow of

upwelled waters before the latter is subducted under the

EAC as shown on the schematic in Fig. 11a, adapted

from Roughan and Middleton (2004).

Through amodeling study, Oke andMiddleton (2000)

showed that the magnitude of the southward transport

of the EAC influenced the amount of cold water up-

welled to the surface. Here, we show from observations

that the intensity of the onshore bottom flow is actually

proportional to the southward current’s magnitude and

similar up- and downstream of the separation point. The

only mooring where this relationship is not evident is

SYD100, which appears to be influenced by the complex

local topography. The southward current’s intensity also

acts on the bottom cross-shelf temperature gradient.

The relationship is linear with a very similar slope for all

locations. It is shown that this bottom temperature

gradient is driven by the onshore bottom flow bringing

cold slope water into the shelf. This average relation-

ship has been quantified from the 2-yr observations. On

the eastern coast of Australia, this implies that around

Smoky Cape (;318S), where the shelf is the narrowest

(16 km) and the EAC reaches speeds of 2 m s21

(Roughan and Middleton 2002), the midshelf bottom

temperature would be 5.58–78C colder than 10 km off-

shore at the same depth.

The occurrence frequency of these bottom slope water

intrusions can also be estimated.Considering an observed

depth-averaged current of 0.3 m s21 at midshelf inducing

a strong enough bottom stress (Fig. 9), the process would

occur roughly 34% and 20% of the time (Fig. 4a), based

on the two years of observations up- and downstream of

the EAC separation point, respectively. The resulting

onshore BBL flow brings slope water, colder by at least

18–1.58C onto the midshelf as compared to the same

depth 10 km offshore. This result is comparable to the

Gulf Stream region where Castelao (2011) estimated

the occurrence of bottom intrusions at midshelf up to

35% of the time in summer. The outcropping of these

water masses was then shown to be much reduced and

related to upwelling-favorable winds. In this study, we

also evidenced a significant intensification of the bot-

tom cross-shelf velocity and the isotherm uplift when

an upwelling-favorable wind blows simultaneously. In

this case, the upwelled water is transported offshore in a

wind-driven Ekman surface flow as shown on the sche-

matic in Fig. 11b.

Unfortunately, the limited temperature observations

at the surface and on the inner shelf did not allow us to

quantify the occurrence of this outcropping process.

Furthermore, while the dynamics of current-driven up-

welling in the BBL are shown to be very similar up- and

downstream of the separation point, the upwelled cold

water can be more or less advected along the coast when

uplifted, depending on the shelf width and current

strength (Oke and Middleton 2001).

Quantifying the occurrence of current- versus wind-

driven upwelling is problematic, as the two processes

can interact and have different impacts on the water

column. Nevertheless, the dominant process upstream

of the EAC separation point appears to be current

related while downstream both processes are expec-

ted to be important, in agreement with the results of

McClean-Padman and Padman (1991) and Roughan

and Middleton (2002).

FIG. 11. (a) Schematic cross-shelf representation of current-driven upwelling in the SouthernHemisphere adapted

from Roughan and Middleton (2004). The average along-shelf geostrophic current is y, Mx is the mass transport

through the bottom boundary layer of thickness d, and Wc is the current-driven uplift. The representation of the

upwelled water subduction is added (light gray arrows) to show the results of this study. (b) Schematic representation

of simultaneous current- and wind-driven upwelling where tsy is the along-shelf wind stress.


To evaluate the impact of these processes, the most

relevant proxy would be related to the resultant biological

productivity. The strong uplifts are expected to have sig-

nificant impacts on the biology as these water masses are

very rich in nutrients. Roughan and Middleton (2002)

suggested a higher nutrient response for current encroach-

ment than wind-driven upwelling from two short-term

hydrographic surveys. This result may vary depending on

the location and the time frame considered.Nevertheless,

the importance of current-driven upwelling along the

eastern coast of Australia for the supply of nutrients to

the euphotic zone, and hence for primary production, is

undeniable.

Acknowledgments. We are grateful for the support of

our partners New South Wales (NSW) Office of Envi-

ronment and Heritage, Oceanographic Field Services,

Connell Wagner Consulting, Manly Hydraulics Labo-

ratory, and Sydney Water Corporation. The Integrated

Marine Observing System is supported by the Australian

Government through theNational CollaborativeResearch

Infrastructure Strategy and the Super Science Initiative.

Data from the ocean reference station (ORS065) were

provided by SydneyWater Corporation. Financial support

is partially provided by a grant from the NSW Office of

Science and Medical Research. We specially thank Linda

Armbrecht for providing the CTD data and Julie Wood

and Vincent Rossi for the insightful discussions.

REFERENCES

Brown,W. S., N. R. Pettigrew, and J. D. Irish, 1985: The Nantucket

shoals flux experiment (NSFE79). Part II: The structure and

variability of across-shelf pressure gradients. J. Phys. Oceanogr.,

15, 749–771.Castelao, R., 2011: Intrusions of Gulf Streamwaters onto the South

Atlantic Bight shelf. J. Geophys. Res., 116,C10011, doi:10.1029/

2011JC007178.

Cresswell, G., 1994: Nutrient enrichment of the Sydney continental

shelf. Aust. J. Mar. Freshwater Res., 45, 677–691.

Dever, E. P., 1997: Wind-forced cross-shelf circulation on the

Northern California Shelf. J. Phys. Oceanogr., 27, 1566–

1580.

Dzwonkowski, B.,K. Park,H.K.Ha,W.M.Graham, F. J.Hernandez,

and S. P. Powers, 2011a: Hydrographic variability on a coastal

shelf directly influenced by estuarine outflow. Cont. Shelf Res.,

31, 939–950, doi:10.1016/j.csr.2011.03.001.

——, ——, and L. Jiang, 2011b: Subtidal across-shelf velocity

structure and surface transport effectiveness on the Alabama

shelf of the northeastern Gulf of Mexico. J. Geophys. Res.,

116, C10012, doi:10.1029/2011JC007188.

Ekman, V., 1905: On the influence of the earth’s rotation on ocean-

currents. Ark. Mat. Astron. Fys, 2, 153.

Everett, J. D., M. E. Baird, P. R. Oke, and I. M. Suthers, 2012:

An avenue of eddies: Quantifying the biophysical properties

of mesoscale eddies in the Tasman Sea. Geophys. Res. Lett.,

39, L16608, doi:10.1029/2012GL053091.

Fewings, M., S. J. Lentz, and J. Fredericks, 2008: Observations of

cross-shelf flow driven by cross-shelf winds on the inner Con-

tinental Shelf. J. Phys. Oceanogr., 38, 2358–2378.

Gibbs, M. T., and J. H. Middleton, 1997: Barotropic and baroclinic

tides on the Sydney continental shelf. Cont. Shelf Res., 17,

1005–1027, doi:10.1016/S0278-4343(97)00004-6.

——, ——, and P. Marchesiello, 1998: Baroclinic response of

Sydney Shelf waters to local wind and deep ocean forcing.

J. Phys. Oceanogr., 28, 178–190.

——, P.Marchesiello, and J. H.Middleton, 2000: Observations and

simulations of a transient shelfbreak front over the narrow

shelf at Sydney, southeastern Australia. Cont. Shelf Res., 20,

763–784, doi:10.1016/S0278-4343(99)00090-4.

Gill, A., 1982:Atmosphere-OceanDynamics. Academic Press, 662 pp.

Godfrey, G., J. Cresswell, T. Golding, A. Pearce, and R. Boyd,

1980: The separation of the East Australian Current. J. Phys.

Oceanogr., 10, 430–440.

Hyun, K. H., and R. He, 2010: Coastal upwelling in the South

Atlantic Bight: A revisit of the 2003 cold event using long term

observations and model hindcast solutions. J. Mar. Syst., 83,

1–13, doi:10.1016/j.jmarsys.2010.05.014.

Lentz, S. J., 2001: The influence of stratification on the wind-driven

cross-shelf circulation over the North Carolina Shelf. J. Phys.

Oceanogr., 31, 2749–2760.

——, 2008: Observations and a model of the mean circulation over

the Middle Atlantic Bight Continental Shelf. J. Phys. Ocean-

ogr., 38, 1203–1221.

——, and D. C. Chapman, 2004: The importance of nonlinear cross-

shelf momentum flux during wind-driven coastal upwelling.

J. Phys. Oceanogr., 34, 2444–2457.——, R. T. Guza, S. Elgar, F. Feddersen, and T. H. C. Herbers,

1999: Momentum balances on the North Carolina inner shelf.

J. Geophys. Res., 104 (C8), 18 205–18 226.

Liu, Y., and R. H. Weisberg, 2005: Momentum balance diagnoses

for the West Florida Shelf. Cont. Shelf Res., 25, 2054–2074,

doi:10.1016/j.csr.2005.03.004.

Malcolm, H. A., P. L. Davies, A. Jordan, and S. D. Smith, 2011:

Variation in sea temperature and the East Australian Current

in the solitary islands region between 2001 and 2008.Deep-Sea

Res. II, 58, 616–627.Mata, M., S. Wijffels, J. Church, and M. Tomczak, 2006: Eddy

shedding and energy conversions in the East Australian Cur-

rent. J. Geophys. Res., 111, C09034, doi:10.1029/2006JC003592.

McClean-Padman, J., and L. Padman, 1991: Summer upwelling on

the Sydney inner continental shelf: The relative roles of local

wind forcing and mesoscale eddy encroachment. Cont. Shelf

Res., 11, 321–345, doi:10.1016/0278-4343(91)90025-2.

Middleton, J. H., D. Cox, and P. Tate, 1996: The oceanography of

the Sydney region.Mar. Pollut. Bull., 33, 124–131, doi:10.1016/

S0025-326X(96)00170-1.

Oke, P. R., and J. H. Middleton, 2000: Topographically induced

upwelling off Eastern Australia. J. Phys. Oceanogr., 30, 512–531.

——, and ——, 2001: Nutrient enrichment off Port Stephens: The

role of the East Australian Current. Cont. Shelf Res., 21 (6–7),

587–606, doi:10.1016/S0278-4343(00)00127-8.

——, and D. A. Griffin, 2011: The cold-core eddy and strong up-

welling off the coast of New South Wales in early 2007.Deep-

Sea Res. II, 58, 574–591, doi:10.1016/j.dsr2.2010.06.006.

——, and P. Sakov, 2012: Assessing the footprint of a regional

ocean observing system. J. Mar. Syst., 105–108, 30–51.——, M. England, and J. Middleton, 2003: On the dynamics of an

observed thermal front off Central Eastern Australia. J. Geo-

phys. Res., 108, 3106, doi:10.1029/2002JC001370.


Ridgway, K., and J. Dunn, 2003: Mesoscale structure of the mean

East Australian Current system and its relationship with to-

pography. Prog. Oceanogr., 56, 189–222.

Rosenfeld, L., 1983: CODE-1: Moored array and large-scale data

report.WoodsHoleOceanographic InstitutionTech.Rep. 83–

23, 186 pp.

Roughan, M., and J. H. Middleton, 2002: A comparison of observed

upwelling mechanisms off the east coast of Australia. Cont.

Shelf Res., 22, 2551–2572, doi:10.1016/S0278-4343(02)00101-2.

——, and ——, 2004: On the East Australian Current: Variability,

encroachment, and upwelling. J. Geophys. Res., 109, C07003,

doi:10.1029/2003JC001833.

——, and B. Morris, 2011: Using high-resolution ocean time series

data to give context to long term hydrographic sampling off

Port Hacking, NSW, Australia. Proc. Oceans 2011, Kona, HI,

IEEE, 1–4.

——, P. R. Oke, and J. H. Middleton, 2003: A modelling study

of the climatological current field and the trajectories of

upwelled particles in the East Australian Current. J. Phys.

Oceanogr., 33, 2551–2564.

——, N. Garfield, J. Largier, E. Dever, C. Dorman, D. Peterson,

and J. Dorman, 2006: Transport and retention in an upwelling

region: The role of across-shelf structure.Deep-Sea Res. II, 53(25–26), 2931–2955, doi:10.1016/j.dsr2.2006.07.015.

——, B. D. Morris, and I. M. Suthers, 2010: NSW-IMOS: An in-

tegrated marine observing system for southeastern Australia.

IOP Conf. Ser.: Earth Environ. Sci., 11, 012030, doi:10.1088/

1755-1315/11/1/012030.

——, H. S. Macdonald, M. E. Baird, and T. M. Glasby, 2011:

Modelling coastal connectivity in a Western Boundary Cur-

rent: Seasonal and inter-annual variability.Deep-Sea Res., 58,

628–644, doi:10.1016/j.dsr2.2010.06.004.

Schumann, E. H., 1986: The bottom boundary layer inshore of the

Agulhas Current off Natal in August 1975. S. Afr. J. Mar. Sci.,

4, 93–102, doi:10.2989/025776186784461882.

Shearman, R. K., and S. J. Lentz, 2003: Dynamics of mean and

subtidal flow on the New England Shelf. J. Geophys. Res., 108,

3281, doi:10.1029/2002JC001417.

Tranter, D. J., D. J. Carpenter, and G. S. Leech, 1986: The

coastal enrichment effect of the East Australian Current

eddy field. Deep-Sea Res., 33, 1705–1728, doi:10.1016/

0198-0149(86)90075-0.

Wilkin, J., and W. Zhang, 2007: Modes of mesoscale sea surface

height and temperature variability in the East Australian Cur-

rent. J. Geophys. Res., 112, C01013, doi:10.1029/2006JC003590.

Wood, J., M. Roughan, and P. Tate, 2012: Finding a proxy for

wind stress over the coastal ocean.Mar. Freshwater Res., 63,528–544.


cross-shelf dynamics in a western boundary current regime: implications ...€¦ · cross-shelf...

Documents